In nucleic acid sequencing and analysis, there is a growing emphasis on the use of high density arrays of immobilized nucleic acid probes. Such arrays can be prepared by massively parallel schemes, e.g., using the selective photomask techniques described in U.S. Pat. No. 5,143,854. Arrays constructed in this manner are typically formed in a planar area of between about 4-100 mm2, and can have densities of up to several hundred thousand or more distinct array members per cm2.
In use, an array surface is contacted with one or more analytes under conditions that promote specific, high-affinity binding of the analyte molecules to one or more of the array members (probes). The goal of the procedure is to identify one or more position-addressable members of the array which bind to the analyte as a method of detecting analyte molecule(s). Typically, the analyte is labeled with a detectable label such as a fluorescent tag, to indicate the one or more array regions where analyte binding to the array occurs. A variety of biological and/or chemical compounds have been used as hybridization probes in the above-described arrays. See, generally, Wetmur, J. (1991) Crit Rev Biochem and Mol Bio 26:227.
For example, such arrays can be used to perform nucleic acid hybridization assays. Generally, in such a hybridization assay, labeled single-stranded analyte nucleic acid (e.g. polynucleotide target) is hybridized to a complementary single-stranded nucleic acid probe. The complementary nucleic acid probe binds the labeled target and the presence of the target polynucleotide of interest is detected.
A common drawback of nucleic acid hybridization assays is the presence of signals which are generated due to an undesirable interaction of various components used in a given assay, i.e. signal generated by entities other than due to hybridization of the analyte and the specified complementary probes, such as signal generated from (i) the reporter, i.e. a signal arising from the label itself when it is not attached to the target, such as a signal generated from a fluorescent dye used in labeling the target; (ii) the non-reporter, i.e. a signal generated from the substrate or other assay components, and (iii) signal generated due to non-specific binding of probes to labeled entities other than their specific target molecules, i.e. binding not related to hybridization of the analyte and the complementary probes. Background signal generated from any of these mechanisms will add to the total signal measured. Uncorrected signal containing background signal results in an overestimation of the “real” signal, which can lead to “false positive” results. Thus, the background signal needs to be estimated accurately and subtracted from the total signal of a hybridization assay to yield the “real” signal.
However, accurate estimation of the background signal is complicated. Underestimation of the background signal will result in an overestimation of the “real” signal, which can yield “false positive” results. Conversely, overestimation of the background signal will result in an underestimation of the “real” signal, which can yield “false negative” results. Thus, background overestimation will negatively impact the lowest concentration of the target that can be reliably detected. An accurate estimate of the background signal is thus needed to generate accurate results.
A common approach to correcting background signal in arrays is to evaluate the portion of the array that is outside of the probe features. However, the background correction problem is particularly complex for measurements made using arrays of nucleic acid hybridization probes, because background may vary as a function of location on the surface. Furthermore, the local properties of the surface that contains bound nucleic acid probes may be very different from the surrounding surface that does not contain bound probes. The “local background signal” is the signal generated from the portion of the array outside of the probe feature area. The signal from the local background immediately adjacent to a given feature is subtracted from the total signal of that feature to correct for background and to yield the “real” signal. Alternatively, the local background signal from the entire array can be evaluated and a single value (e.g. an average local background signal or the minimum local background signal) can be calculated to correct all features of that array. This is referred to as the “global background signal”. The choice of an appropriate background correction method depends critically upon which of these two influences, i.e., local background or modification of surface properties by covalently bound nucleic acid probes, is judged to most strongly influence background signal in the array regions containing covalently bound probe molecules.
The use of local or global background correction methods are problematic. The properties of the array surface outside the features may differ from the properties of the array surface within the features. These differences can result in different levels of non-reporter signal or different levels of reporter non-specific binding. Thus, the observed signal from the local background or estimated from a global background calculation may result in an inaccurate estimation of the background signal within the feature. Additionally, the probes themselves may generate a portion of the background signal. For example, the bases or phosphodiester linkages of the probes may (i) produce non-reporter signal, (ii) bind to components which produce non-reporter signal, or (iii) non-specifically bind the reporter. Therefore, in these cases, using local background will underestimate the true background signal that should be subtracted.
Representative methods for resolving the problem of interfering background signals in nucleic acid hybridization assays are described in U.S. Pat. Nos. 4,868,105; 5,124,246; 5,563,034; and 5,681,702; WO 98/24933; Chen Y., et al., Journal of Biomedical Optics (1997) 2:364-374; and DeRisi J. L. et al. (1997) Science 278:680-686. Existing methods generally correct for background signal by subtracting either the local or global background. However, these methods do not involve surface-bound nucleic acid probes, and in some cases background estimates obtained from local or global sampling of nonprobe regions overestimate background in regions that contain probes. Background overestimation negatively impacts the lowest dose of the target that can be reliably detected by an array involving a nucleic acid hybridization assay.
Therefore, there is a need for reliable methods for estimating background signal from probe-containing regions in hybridization arrays.